• PDF / 349,458 Bytes
• 6 Pages / 420.48 x 639 pts Page_size
• 81 Downloads / 187 Views

DOWNLOAD

REPORT

---

THE EFFECT OF VACANCIES GROWN INTO SILICON ON GOLD DIFFUSION R.K. Graupner *, J.A. Van Vechten**, P. Harwood***, and T.K. Monson** * Komatsu Silicon U.S.A., Santa Clara, CA 95051 ** Oregon State University, Corvallis, OR 97331-3211 *** Wacker Siltronic, Portland, OR 97283-0180 ABSTRACT We propose a generalized model for gold diffusion in silicon based on the effect of the high concentrations of vacancies and vacancy complexes in the as-grown silicon. The monovacancy profiles calculated using this model are identical to the substitutional gold profiles calculated using the kick-out model. We deposited Au on commercial float zone Si in a vacuum system after the Si had reached the diffusion temperature (1233 K) and had been annealed in various ways. Contrary to previously published reports, we find the electrically active Au with a nearly one-sided profile when the Au is deposited on samples which were preannealed in vacuum. We conclude that annealed silicon surfaces lack the imperfections needed to make them effective sources or sinks for vacancies or self-interstitials. We propose that this can cause a high degree of supersaturation in the asgrown silicon crystal since the point defects cannot annihilate at the surfaces to maintain equilibrium as the crystal is cooled. INTRODUCTION The diffusion of gold into silicon has been the subject of extensive experiment, theory and controversy [1,2,3,4,5,6]. Previously reported experiments show a distinctive U-shaped profile for electrically active, substitutional gold, Ausi, which results even when the source is on one side only and precautions are taken to avoid surface diffusion. It is generally agreed that the Au diffuses rapidly as an interstitial, Au,, and that the concentration of Au., is orders of magnitude less than the equilibrium concentration of Ausi. There is controversy concerning how Aui becomes Ausi with a symmetric profile from a one-sided source. Initially a Frank-Turnbull (FT) or dissociative mechanism (1) was invoked to explain this profile. In this mechanism vacancies, V, generated at the surfaces diffuse into the bulk and combine with Aui to form Ausi.

Au +V-fAAus,

(1)

The FT mechanism with the assumption that the vacancies are only generated at the surfaces has not yet been shown to explain the profiles observed in dislocation-free silicon. Van Vechten and Zhang have shown through direct Monte Carlo simulation that the FT mechanism cannot produce the reported nearly symmetric profile from a strictly one-sided source starting with a perfect sample. These simulations, to be published elsewhere, are two-sided but not symmetric and differ Mat. Res. Soc. Symp. Proc. Vol. 262. 01992 Materials Research Society

538

significantly from the reported profiles. Others claim that the profile results from a "kick-out" (KO) and is determined by the creation of silicon selfmechanism and their the bulk via reaction (2) Si., in interstitials, subsequent out-diffusion and annihilation at the wafer surfaces.

(2)

Au, Aus,+SI,

Both of these models assume that the sam